To test whether JAG1 induces osteoblast commitment and differentiation in human osteoblast-like primary cells, we derived human bone osteoblast-like (HBO) cell lines (HBO1, 2, 3, 4, 5, 6, and 7) from seven healthy pediatric human bone samples, as described (Lockyer et al., 2007; Figure 1—figure supplement 1). HBO cells were then treated with Fc-Dynabeads (Fc-bds) (5.7 μM) as a negative control, since JAG1 is a chimeric recombinant protein with an Fc-portion, and JAG1-Dynabeads (JAG1-bds) (5.7 μM), in the presence of osteogenic media. On day 21, the cells were stained for mineralization using Alizarin Red S stain (Figure 1A, B). We observed significantly increased mineralization in JAG1-bds-treated samples compared to growth media-, osteogenic media-, and Fc-bds-treated (p = 0.0107) cells. Additionally, PCR analysis of HBO1 cells from a repeat experiment collected at days 7, 14, and 21 showed significantly increased expression of osteogenic genes with JAG1-bds stimulation (Figure 1C). ALPL was significantly expressed at day 7, with a 3.5-fold increase (p = 0.0004) compared to HBO1 cells grown in growth media. In contrast, significant expression levels of COL1A1 and BGLAP were observed at 14 days, with a 5.1-fold increase (p = 0.0021) of COL1A1 and a 12.3-fold increase (0.0002) of BGLAP when compared to growth media conditions. Interestingly, while some mineralization is observed in the osteogenic media and Fc-bds (Figure 1A) conditions, there were no significant increases in osteogenic gene expression (Figure 1C). Expression of RUNX2 and SP7 was not significantly altered across all conditions and time points (not shown). In preparation for in vivo studies, HBO cells were incorporated into JAG1-Dynabead-PEG-4MAL and grown in vitro followed by Alizarin Red S staining to assess mineralization (Figure 1—figure supplement 2). These data indicate that JAG1 can induce osteoblast commitment, differentiation, and mineralization of pediatric HBO cells.

Since we observed an induction of osteoblast commitment and differentiation in pediatric HBO cells in vitro, we next assessed whether co-delivery of JAG1-presenting hydrogels with pediatric HBO cells act similarly in murine CF defects. We recently reported that JAG1 can induce murine CNC cell osteoblast commitment and repair cranial bone defects in vivo (Kamalakar et al., 2021). To establish a more translatable use of JAG1 in treating CF bone loss, we assessed whether JAG1 can stimulate human cells (pediatric HBO) to facilitate bone regeneration in NOD-SCID mice, to prevent graft rejection of the HBO cells. JAG1-Dynabead-PEG-4MAL hydrogels also encapsulating pediatric HBO cells obtained from three separate donors were implanted in critically sized parietal bone defects (4 mm) in NOD-SCID mice (n = 4–6 per donor, 13–15 total) (Figure 2—figure supplement 1). The volume of bone regenerated by the JAG1-PEG-4MAL-pediatric HBO hydrogel with or without DAPT, an inhibitor of NOTCH canonical signaling, was measured using micro computed tomography (μCT), and compared to HBO cells alone, Fc-bds (20 μM) + BMP2 (2.5 μM) treatments. To maintain continuous osteogenic induction, additional JAG1-PEG-4MAL-pediatric HBO cells and control hydrogels were injected into the defects transcutaneously at week 4. After 8 weeks, we quantified differences in bone volume (BV) within the cranial defect using μCT analysis (Figure 2A, B). We observed that there was minimal bone regenerated in mice treated with cells alone. As expected, BMP2 significantly increased regenerated BV (p = 0.0002) compared to the cells alone group. The BV regenerated by JAG1-bds in the absence (fold change: 1.5) and presence of DAPT (fold change:1.6) was significantly higher compared to the cells alone treatment group (p = 0.0092 and 0.0021, respectively). There was no sex-based difference in regenerated BV. An initial pilot study also demonstrated no difference in bone regeneration between an Empty Defect model (no HBO cells) and a Cells Alone group (Figure 2—figure supplement 2). In Figure 2C, paraffin sections of mouse skulls were stained with Masson trichrome stain, as described in methods under histology. Qualitative assessments of the stained sections revealed increased collagen (blue color) in samples obtained from mice treated with BMP2, as expected, and with JAG1-bds and JAG1-bds + DAPT compared to the mice treated with cells alone. To validate the results observed in the samples stained with Masson trichrome, we performed immunohistochemical 3,3'-diaminobenzidine (DAB) staining with additional sections from the same mice using rabbit IgG monoclonal anti-COL1A1 antibody. These results corroborate what was revealed with the Masson trichrome staining, that mice treated with cells alone had less collagen production (Figure 2—figure supplement 3). This suggests that JAG1 can be used as a bone regenerative therapy where JAG1 induces bone regeneration independently of NOTCH canonical signaling in human cells.

We previously found that JAG1 can activate a NOTCH non-canonical JAK2–STAT5 signaling pathway in mouse CNC cells, which stimulated expression of osteoblast genes (Runx2 and Bglap (Osteocalcin)) as well as osteoblast commitment and proliferation (Kamalakar et al., 2019; Kamalakar et al., 2021). Thus, we asked if JAG1 would have similar effects for a non-canonical NOTCH pathway on HBO cells. To evaluate this, we cultured HBO cells from a single donor in triplicate and conditioned for 24 hr with vehicle, DAPT, JAG1-bds, or both. Comparison of the JAG1-bds and JAG1-bds + DAPT conditions revealed clusters of genes that were up- or downregulated by JAG1 stimulation and remained with inhibition of NOTCH, that is, defining the non-canonical pathway (Figure 3A). Analysis of differentially expressed genes (DEGs) in JAG1-bds and JAG1-bds + DAPT groups compared to no treatment revealed a total of 448 upregulated genes and 435 downregulated genes in the non-canonical pathway (Figure 3B, Supplementary file 1). These include upregulation of genes involved in osteoblast commitment (RUNX2), matrix remodeling (MMP3), and diverse cytokines and chemokines (CCL5, CXCL1, and CXCL6) as part of the non-canonical pathway (Figure 3C). Gene ontology (GO) analysis of the upregulated genes in the non-canonical pathway revealed significant over-representation of GO terms associated with RUNX2, cytokine signaling, Nuclear factor kappa B (NF-κβ), and cell cycle (Figure 3D). More interestingly, the PIP3 activating AKT signaling pathway was upregulated by JAG1-bds treatment, suggesting that JAG1 can activate NOTCH non-canonical signals via the AKT pathway.

Collectively, these data suggest that JAG1 has a profound NOTCH non-canonical effect on HBO cells that stimulates HBO cell-osteoblast commitment and differentiation leading to HBO-cell-induced bone formation.

Having observed that JAG1-induced osteoblast commitment of murine CNC cells via a NOTCH non-canonical pathway (JAG1–JAK2), we sought to determine whether JAG1 activated NOTCH non-canonical pathways and targets in the pediatric HBO cells. Serum-starved pediatric HBO cells were subsequently treated with JAG1-bds (5.7 μM) with or without DAPT (15 μM), to block the NOTCH canonical pathway, as a time course stimulation for 5, 10, 15, and 30 min. Phosphorylation levels for signaling molecules were assessed via Luminex-based multiplex assays. We observed significantly increased phosphorylation of multiple signaling molecules, including STAT5, AKT, P38, JNK, RELA, and p70 S6K in JAG1-bds-treated cells, even in the presence of DAPT (Figure 4). We have previously shown that JAG1 induced the phosphorylation of STAT5 during CNC cell differentiation to osteoblasts (Kamalakar et al., 2021). Prior studies emphasize the importance of non-canonical signaling, crosstalk between NOTCH, and other cellular signaling mechanisms. For example, the WNT pathway cross-talks with the NOTCH canonical pathway during vascular morphogenesis (Caliceti et al., 2014). Furthermore, previous reports have shown that STAT5 is essential for AKT–p70 S6K activity during lymphocyte proliferation in patients with leukemias and lymphomas (Lockyer et al., 2007). Similarly, the P38 pathway has been shown to activate the mammalian target of rapamycin (mTOR)–p70 S6K pathway during oxidative stress in mouse embryonic fibroblast cells which culminates in upregulation of antioxidative enzymes that assist in reactive oxygen species removal and thereafter increase cell survival (Gutiérrez-Uzquiza et al., 2012). Also shown previously, JNK phosphorylates p70 S6K to induce osteoblast proliferation and differentiation of MC3T3 cells, and can crosstalk in other physiological systems, for example, during hepatocyte proliferation (Bouxsein et al., 2010; Love et al., 2014; Zhao et al., 2014). A study by Miwa et al., in 2012, showed that AKT–mTOR–p70 S6K, extracellular signal-regulated kinases (ERK), and NF-κB were involved together in proliferation of osteosarcoma cells and these pathways could be inhibited by caffeine thereby decreasing tumor burden (Miwa et al., 2012). We observe in our results that many of the pathways activated by JAG1 in the pediatric HBO cells lead to the phosphorylation of p70 S6K, downstream. These findings implicating phospho-protein signaling pathways, especially NF-κB and AKT (downstream of phosphatase and tensin homolog (PTEN)) are consistent with our transcriptional profiling (Figure 3) and collectively implicate p70 S6K as a potential downstream contributor to JAG1-induced, pediatric HBO cell-mediated bone regeneration. Therefore, we proceeded to determine if p70 S6K is an essential downstream target of the JAG1-induced NOTCH non-canonical signaling in pediatric HBO cells (Figure 5).

As shown previously, we measured significantly increased alkaline phosphatase production as well as mineralization in JAG1-bds-treated samples compared to all other treatments. We also showed an increase in the phosphorylation of p70 S6K. According to multiple reports in literature, phosphorylation of p70 S6K occurs downstream of multiple pathways (AKT, JAK–STAT, and P38) that we identified in JAG1-induced pediatric HBO cells (Lockyer et al., 2007; Gutiérrez-Uzquiza et al., 2012; Miwa et al., 2012). Thus, we next tested whether the phosphorylation of p70 S6K is an essential downstream target of JAG1-NOTCH during HBO cell mineralization. HBO cells were treated with growth media alone, osteogenic media alone, or Fc-bds (5.7 μM) as negative controls, and JAG1-bds (5.7 μM) with or without S6K-18, an inhibitor of phosphorylated p70 S6K (Ye et al., 2011), in the presence of osteogenic media. On day 21, the cells were stained for mineralization using Alizarin Red S stain. As shown in Figure 5, we observed that negative controls did not induce mineralization of the HBO cells while JAG1-bds induced significantly higher levels of mineralization, as expected. S6K-18-treated samples showed partial inhibition of 50.2% of the JAG1-bds-induced mineralization (p = 0.0015) and not 100%, possibly because the NOTCH canonical pathway was not inhibited in these samples. Inhibition of JAG1-bds-induced mineralization by S6K-18 treatment alone was significant compared to mineralization induced by JAG1-bds alone suggesting that the phosphorylation of p70 S6K is an essential event downstream of JAG1-NOTCH in JAG1-stimulated HBO cells. Additionally, inhibition of JAG1-bds-induced mineralization was observed with DAPT, an inhibitor of canonical NOTCH signaling (Figure 5—figure supplement 1). Subsequent repeat experiments involved collecting the cells on days 9, 14, and 21 for qRT-PCR analysis. While inhibition of NOTCH and p70 S6K decreased mineralization in our mineralization assay, there are no statistically significant changes in gene expression for ALPL, COL1A1, or BGLAP (Figure 5—figure supplement 2). These results suggest that the HBO cells phenotypes are maturing into osteocytes and that inhibiting p70 S6K hinders the cellular ability to mineralize but not the cell phenotype progression.

HBO cell lines were derived from healthy fibulas of seven pediatric subjects under appropriate Institutional Review Board approval. Segments of the fibulas were digested using collagenase A (Roche, 10103578001) used at 0.1 mg/ml treatment for 40 min with replacement of collagenase A midway at 20 min and then digested with 0.2 mg/ml of collagenase A for 60 min at 37°C with intermittent shaking. The segments of bone were maintained in Dulbecco’s modified Eagle medium (DMEM) + Primocin Antimicrobial agent for primary cells (Invivogen, ant-pm-1) + 10% fetal bovine serum (FBS) + 50 µM ascorbic acid (Sigma, 49752) + 10 nM dexamethasone. Media changes were performed every 5 days. Osteoblast-like cells began to grow out of the bone segments by day 10. These cells can be passaged and frozen for storage. HBO cell lines used for experiments were selected based on the ability of the primary cell line to proliferate and mineralize in culture. On addition of Osteogenic media [DMEM + Primocin + 10% FBS + 50 µM ascorbic acid + 10 nM dexamethasone (Sigma, D1756) + 10 mM beta-glycerophosphate disodium (Sigma, G9422)], the cells form mineralized nodules indicating their osteogenic ability as seen in Figure 1—figure supplement 1.

As per the manufacturer’s recommendations, 50 μl (1.5 mg) of Dynabeads Protein G (Invitrogen 10004D) were transferred to a tube, where the beads were separated from the solution using a magnetic tube rack (Bio-Rad, 1614916) and washed once using 200 μl phosphate-buffered saline (PBS) with 0.1% Tween-20 (Fisher, BP337-500). The wash buffer was separated from the beads-Fc complex using the magnetic rack. Recombinant JAG1-Fc (5, 5.7, 10, or 20 µM) (Creative Biomart, JAG1-3138H) or control IgG-Fc fragment (5 or 5.7 µM) (Abcam, ab90285) were diluted in 200 μl PBS with 0.1% Tween-20 and then added to the Dynabeads. The beads plus proteins were incubated at 4°C with rotation for 16 hr. Thereafter, the tubes were placed back on the magnetic rack and the supernatant was removed. The bead–JAG1/Fc complex was resuspended in 200 μl PBS with 0.1% Tween-20 to wash by gentle pipetting. The wash buffer was also separated from the beads–Fc complex using the magnetic rack, and the final suspension of the beads in hydrogels was used as treatment.

HBO cells were seeded at 30,000 cells per well in a 12-well plate, treated (n = 3) with and cultured for 21 days in osteogenic media (DMEM + Primocin + 10% FBS + 50 µM ascorbic acid + 10 nM dexamethasone + 10 mM beta-glycerophosphate) with half feeds every 5 days. Treatments included growth media (DMEM + Primocin + 10% FBS), osteogenic media, Fc-Dynabeads (5.7 μM) in osteogenic media or JAG1-Dynabeads (5.7 μM) in osteogenic media. Inhibitors N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT; Sigma, D5942) (15 µM) and 5-(1,1-dimethylethyl)–2-[[(1H-indazol-5-ylamino)carbonyl]amino]-3-thiophenecarboxylic acid (S6K-18 – inhibitor of p70 ribosomal S6 kinase 1; Selleck Chemicals, S0385) (50 μM) were also added to JAG1-Dynabeads conditions for experiments shown in Figure 5 and Figure 5—figure supplements 1 and 2. For PCR, cells were collected on days 7–9, 14, and 21 (see Methods ‘RNA extraction and qRT-PCR’). For Alizarin Red staining: On day 21, the cells were fixed using 50% ethanol for 15 min at 4°C. The fixed cells were then stained with Alizarin Red S dye (LabChem, LC106002) to detect mineralization. The dye was extracted using a 10% acetic acid solution in water and quantified by measuring the absorbance at 420 nm using a spectrophotometer.

We prepared poly (ethylene glycol) (PEG)-based synthetic hydrogels incorporating cell adhesive peptides in two steps. First, maleimide end-functionalized 20 kDa four-arm PEG macromer (PEG-4MAL, with >95% end-group substitution, Laysan Bio, 4ARM-PEG-MAL-20K) was reacted with a thiol-containing adhesive peptide GRGDSPC (RGD, Genscript, RP20283) in PBS with 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) at pH 7.4 for 1 hr. Then, the RGD-functionalized PEG-4MAL macromers were cross-linked in the presence of HBO cells and JAG1-Dynabeads into a hydrogel by addition of the dithiol protease-cleavable peptide cross-linker GPQ-W (GCRDGPQGIWGQDRCG) (New England Peptides, Inc (NEP) Custom synthesized) (Kamalakar et al., 2019; Phelps et al., 2012). The final gel formulation consisted of 4.0% wt/vol polymer and 1.0 mM RGD.

We performed all in vivo experiments using procedural guidelines with appropriate approvals from the Institutional Animal Care and Use Committee of Emory University (#PROTO201700263). Six- to eight-week-old male and female NOD.Cg-Prkdc^scid/J mice (The Jackson Laboratory, 001303) were used. As shown in Figure 2—figure supplement 1, the surgery site was disinfected and then incisions were made using sterile surgical equipment to expose the parietal bones of the mice. Thereafter, 4 mm defects were created in the parietal bones using a variable speed drill (Aseptico (MicroNX), MAX-88ESP, CL1791023) and sterile circular knives. JAG1 delivery: PEG-4MAL hydrogels (20 µl) loaded with JAG1-Dynabeads without or with 100,000 HBO cells (n = 13–15 for all treatment groups) were placed within the defects created in parietal bones in the NOD-SCID mouse skulls as the first dose. A second dose of the hydrogels encapsulating HBO cells and Dynabead-bound JAG1 were administered as transcutaneous injections during week 4 to continue the bone regenerative action of JAG1. Skulls were then harvested at week 8, fixed using 10% neutral buffered formalin (VWR, 89370-094) and imaged with µCT. HBO primary cell lines 2, 6, and 7 from separate individuals were selected for these experiments based on similar growth and passage characteristics. Inclusion criteria: Survival to end of the experiment. Exclusion criteria: Illness that required euthanasia before the end of the experiment or death.

µCT analyses were conducted according to current guidelines for the assessment of BV within the defects created in mouse calvaria (Bouxsein et al., 2010). Briefly, formalin-fixed skulls were positioned in the µCT tubes with the nose facing the bottom of the tube and imaged in a µCT 40 (Scanco Medical AG, Bassersdorf, Switzerland) using a 36-µm isotropic voxel size in all dimensions. Thereafter, using a consistent and pre-determined threshold of 55 kVp, 145 µA, 8 W and 200 ms integration time for all measurements, three-dimensional reconstructions were created by stacking the regions of interest from ~600 two-dimensional slices consisting of the entire skull and then applying a gray-scale threshold of 150 and Gaussian noise filter (σ  =  0.8, support  =  1.0), a coronal reformatting was done. Thereafter, a circular region of interest encompassing the defect was selected for analysis consisting of transverse CT slices encompassing the entire defect and, new BV was calculated.